Time for OTP to be Renamed?

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Time for OTP to be Renamed?

jocke
I would only use the gen_server iff I needed hot code loading/upgrading. In the real world, and in the majority of cases, that seldom is needed.

IMHO and within my experience of course.

Den 16 feb 2014 10:52 skrev Vance Shipley <vances>:

>
> On Sun, Feb 16, 2014 at 10:40:52AM +0100, Joacim Greben? wrote:
> }? The gen_server takes away all the fun.
>
> It does, true.? If you want to write pure Erlang and still play
> by the (OTP) rules you can use plain_fsm(*) or sys_fsm(+) or
> write it yourself.? Not using OTP behaviours is just harder to
> get right and much less tested.
>
> (+) http://erlang.org/pipermail/erlang-questions/2004-February/011405.html 
> (*) http://erlang.org/pipermail/erlang-questions/2005-April/015226.html 
>
> --
> -Vance

Reply | Threaded
Open this post in threaded view
|

Time for OTP to be Renamed?

Vance Shipley-2
On Sun, Feb 16, 2014 at 11:06:08AM +0100, Joacim Greben? wrote:
}  I would only use the gen_server iff I needed hot code loading/upgrading.
}  In the real world, and in the majority of cases, that seldom is needed.
}  
}  IMHO and within my experience of course.

Lest other, less experienced, readers get the impression that it is that
simple let me just provide this pointer:

   http://www.erlang.org/doc/design_principles/spec_proc.html

To write a program which will work (properly) in an OTP compliant system,
you must receive and handle a number of system messages.  Without OTP
compliance you have no supervision, debugging, release handling, observer
and I don't know what else.  You are free to do so but I wouldn't recommend
this route for anyone else who hasn't mastered the environment.

--
        -Vance